What ChatGPT Computer History Is—and Why It Matters
ChatGPT Computer History is an opt‑in feature in the ChatGPT desktop app for macOS that records your clicks, typing, and app switches across allowed apps and websites, then turns that activity into text memories and a searchable timeline that ChatGPT and Codex can use as context in later conversations. OpenAI launched Computer History as a new feature in its macOS desktop client, appearing in the product changelog on August 13 and rolling out to ChatGPT Pro, Business, and Enterprise users. In other words, OpenAI is doing something bold: turning your daily computer activity into AI‑ready memory while trying not to look like a surveillance system. This is a deliberate pivot away from screenshot logging and toward event‑level tracking that lives, at least initially, on your own machine.
From Screenshot Spying to Event Streams
The most important design choice in Computer History is what it doesn’t record. Instead of grabbing images of your screen as Chronicle did, it listens to interaction events exposed through macOS accessibility features—clicks, typing, keyboard shortcuts, app switches—across the apps and websites you allow. OpenAI is explicitly distancing itself from Recall‑style screenshot surveillance in favor of metadata: what you did, not what every pixel looked like. The desktop app builds an interaction‑event stream and periodically spins up an ephemeral Codex session to summarize that stream into text and store it as local memory files. Those memories are plain‑text Markdown documents saved on your machine under a Codex directory, and you can open, read, and edit them like any other file. This is desktop app tracking by design—but it is tracking you can inspect, not a black box.
Turning Activity Into AI Memory You Can Search
Computer History’s AI memory features are not theoretical; they show up in two concrete places. First, there is a timeline view in Settings that groups summaries by day and time, showing which apps contributed to each entry and letting you jump straight to the underlying memory file in Finder or delete it. Second, those same memories become fuel for more context‑aware answers: ChatGPT and Codex can tell you what you were working on before a break, where a document you viewed earlier lives, or summarize yesterday’s work for a standup without you reconstructing it from scratch. Think of it as an AI that stays in the flow of your desktop life instead of treating every chat as a blank slate. The trade‑off is obvious: more awareness for the AI means more detailed logs of your behavior.
Opt‑In Activity Tracking Privacy: Better, Not Perfect
On paper, Computer History is a textbook example of activity tracking privacy done better than the industry’s screenshot land grab. OpenAI stresses that Computer History does not capture screenshots, screen recordings, microphone input, or system audio, and that private‑mode web browsing is never included. The raw event stream lives in an isolated container on your Mac, is processed on OpenAI’s servers to generate memories, and is deleted after 48 hours. It’s also off by default, and each Pro user has to turn it on manually; Business and Enterprise users need both admin approval and individual opt‑in. This is a meaningful shift: OpenAI is betting that users will volunteer their activity when they see clear benefits and controls, instead of being dragged into full‑screen capture they cannot reasonably audit.
But “better” is not the same as “safe.” The company’s own documentation warns that Computer History files can contain sensitive information, are not encrypted by the feature, and that other programs running under the same macOS user account may be able to access them. In practice, that means the weakest link might be other software on your machine, not OpenAI’s servers. Users do get some power tools: you can choose which apps and websites contribute, pause collection from a menu bar shortcut, and delete your history at any time. Still, if you are uncomfortable with a keylogging and event capture system sitting on your desktop—even a “friendly” one—your skepticism is justified.
From Memory to Automation—and What Comes Next
Computer History is not just about recall; it is about turning your habits into automations. When the system spots repeatable work in the activity stream, a timeline entry can propose transforming that workflow into a reusable skill or scheduled automation that Codex then builds from the recorded steps. One documented example is a daily recap automation suggested after the system notices the user assembling the same summary each morning. This closes a loop in OpenAI’s agent strategy: the system observes your work, remembers it, and then proposes automating it. That is a powerful move, and it pushes Computer History from convenient memory aid into quiet workplace infrastructure. Computer History is currently limited to the ChatGPT desktop app on macOS, with users in certain regions promised access in the coming weeks. It is hard to believe it will stay Mac‑only for long; once an AI can read your computer history, every platform becomes a potential training ground.
